cvfile-haystack 0.3.2

Haystack integration for the .cv open file format.

cvfile-haystack

Haystack 2.x converter component for the .cv open file format.

A .cv file is a PDF/A-3u file carrying a Markdown copy of the same content (plus optional HTML and JSON Resume) as PDF Associated Files. Instead of OCR ing the PDF, this component reads the embedded text payloads directly and emits Haystack Document objects ready for indexing.

Install

pip install cvfile-haystack

Use

from haystack_integrations.components.converters.cvfile import CVFileToDocument

converter = CVFileToDocument()
result = converter.run(sources=["resume.cv"])
documents = result["documents"]

for doc in documents:
    print(doc.meta["payload"], doc.meta["mime_type"], len(doc.content))

You get one Document per textual payload found in the file. The Markdown copy (typically resume.md) is the one flagged with meta["primary"] = True.

Primary only

If you only want the canonical Markdown copy and want to skip language alternates and supplements:

converter = CVFileToDocument(primary_only=True)

Untrusted files

By default the converter runs cvfile.validate() on every source before extracting anything. Files carrying forbidden active content (JavaScript, launch or submit actions, external references), encryption, integrity digest mismatches, or payloads over the spec size cap make run() raise ValueError listing the issue codes. Resumes are classic untrusted input, so keep the default when converting files you did not produce yourself.

converter = CVFileToDocument()              # verify=True (default)
converter = CVFileToDocument(verify=False)  # trusted files only

Pipeline use

from haystack import Pipeline
from haystack.components.embedders import SentenceTransformersDocumentEmbedder
from haystack.components.writers import DocumentWriter
from haystack.document_stores.in_memory import InMemoryDocumentStore
from haystack_integrations.components.converters.cvfile import CVFileToDocument

store = InMemoryDocumentStore()
pipe = Pipeline()
pipe.add_component("read", CVFileToDocument(primary_only=True))
pipe.add_component("embed", SentenceTransformersDocumentEmbedder(model="BAAI/bge-m3"))
pipe.add_component("write", DocumentWriter(document_store=store))
pipe.connect("read.documents", "embed.documents")
pipe.connect("embed.documents", "write.documents")

pipe.run({"read": {"sources": ["resumes/jane.cv", "resumes/john.cv"]}})

Metadata fields

Key	Description
source	The file path (or stream name) the document came from
payload	Name of the embedded file (e.g. resume.md)
mime_type	MIME of the payload (text/markdown, text/html, application/json)
relationship	PDF Associated Files relationship (Alternative for primary alternates)
language	BCP 47 language tag for this payload
primary	True for the payload declared as primary in the file's XMP metadata
cv_version	Version of the .cv spec the file conforms to
cv_generator	Tool that produced the file, if recorded

License

Apache-2.0.